Pancreatic ductal adenocarcinoma (PDAC) is an almost uniformly fatal malignancy that often presents clinically as late stage invasive and metastatic disease. Both cancer cell intrinsic and extrinsic factors contribute to pancreatic cancer initiation and progression but most steps remain poorly understood at the molecular level. Genetically engineered mouse models of human PDAC have been important tools to understand the early stages of pancreatic cancer development, many aspects of basic cancer biology, and evaluate new treatment strategies. Assessing gene function using conventional mouse models of pancreatic cancer is expensive, labor intensive, and slow with only minor efforts towards generating more accurate, rapid, and flexible alternatives. Furthermore, despite the clear importance of altering genes of interest within established pancreatic cancer this remains a major largely unanswered challenge. In this high-risk high-reward proposal, we document preliminary data and outline systems that will allow rapid functional interrogation of candidate genes in vivo at any stage of pancreatic cancer progression. We will generate flexible virus-based autochthonous mouse models of pancreatic cancer using a method of tumor induction based on retrograde pancreatic duct injection of viral Cre-expressing vectors. This will allow the rapid interrogation of gene function during pancreatic cancer development in vivo. Incorporation of lentiviral vectors containing a cDNA or shRNA targeting a gene of interest will enable genetic manipulation of PDAC in vivo no more difficult than altering a cell line in culture. To allow streamlined genomic alterations in mouse models of human pancreatic cancer we have also generated a mouse with Cre-regulatable Cas9 expression. In combination with the conventional conditional KrasG12D and tumor suppressor alleles this will allow lentiviral vectors carrying Cre and a guideRNA to induce CRISPR- mediated genome engineering in pancreatic cancer. Finally, we will combine viral-FLP induced pancreatic cancer initiation with secondary viral-Cre infection to enable temporally separated genetic and genomic alterations. These methods will greatly simplify and accelerate investigation of gene function in vivo and provide a path towards multiplexed combinatory genetic manipulations as well as pooled gain- and loss-of- function screens in pancreatic cancer in vivo. We have extensive expertise using genetic-engineering approaches to generate disease models, a growing effort in PDAC, considerable experience using other viral induced tumors models, and are well integrated within the Stanford pancreatic cancer basic science and clinical research community. PDAC patients have a 5 year survival rate of ~5% underscoring the need for truly novel approaches to accelerate the molecular characterization of this disease. With growing interest in the interaction of pancreatic cancer cells with the immune system and other stromal components during each step of carcinogenesis as well as the relative void of mechanistic knowledge related to metastatic progression, our high-risk project has the potential to benefit many pancreatic cancer investigators.